diff options
| author | Kawrakow <iwankawrakow@gmail.com> | 2025-06-18 16:20:54 +0300 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2025-06-18 16:20:54 +0300 |
| commit | d85c64428e0598f2617a352aec9960242af45652 (patch) | |
| tree | 66bd9be9d2ea315917b36adf1ac4cd24b9ba7a97 /ggml/src/ggml-cuda/common.cuh | |
| parent | c410cc72bbfcbdef9ce552b425ab7abbeb250dff (diff) | |
New IQ2_KT, IQ3_KT and IQ4_KT, V2 (#529)
* New iq4_kt trellis
The new trellis generates int8_t values via
sum_as_uint8_t[(ka * idx + kb) & 0x3f33f3f3f] - 126.
CUDA dequantize works.
AVX2 case Ny > 32 works, and we get 273 t/s for L3-8B.
PPL is on par or even slightly lower than original QTIP trellis.
* Something is not working with the AVX2 dot product
* New iq4_kt: CUDA MMVQ
* New iq4_kt: CUDA MMQ
* For now have only iq4_kt use the new trellis
* Fix iq2_kt that got broken along the way
* New iq4_kt: AVX2 dot product finally works
We get 13.6 t/s vs 8.4 t/s with the f16 trellis and f32 arithmetic.
Still somewhat slower than other quants, but no longer pathetic.
* New iq4_kt: fix vanilla AVX2
* New iq4_kt: NEON implementation
We get very respectable PP-512 = 120 t/s.
TG-128 is pathetic at 5.3 t/s, so 20+% slower than the f16 variant.
* New iq4_kt: slightly faster NEON
* New iq4_kt: slightly faster NEON
* New iq4_kt: faster NEON
We are now at 9.4 t/s, up from 6.6 t/s for the f16 trellis.
* Minor
* New iq4_kt trellis: not working Metal implementation
* Remove the extra 4 bytes of row meta data that is no longer used
* Cleanup
* Adding forgottent file
* Switching iq2_kt to new trellis - CUDA MMQ
* New iq2_kt: CUDA GEMV
* New iq2_kt: AVX2 dequantize
* New iq2_kt: AVX2 GEMM/GEMV
* Adding forgotten file
* New iq2_kt: NEON GEMM/GEMV
* New iq2_kt: slightly faster NEON GEMM
* New iq2_kt: Metal - very slow.
It seems Apple Silicon cannot quickly add 4 8-bit ints.
Or I don't know how to do it - but I didn't find anything
in the Metal Shading Language Specification.
So, performance is quite a bit worse than the original trellis.
* Add missing break
* Trying @louiehelm's multiplier
* CPU
* iq3_kt: use integer trellis + CUDA dequantize and MMVQ
* iq3_kt: MMQ
* iq3_kt: AVX2 GEMM
* iq3_kt: AVX2 GEMV
* The trellis quants now need super-blocks of 256, so we need a check
---------
Co-authored-by: Iwan Kawrakow <iwan.kawrakow@gmail.com>
Diffstat (limited to 'ggml/src/ggml-cuda/common.cuh')
| -rw-r--r-- | ggml/src/ggml-cuda/common.cuh | 14 |
1 files changed, 14 insertions, 0 deletions
diff --git a/ggml/src/ggml-cuda/common.cuh b/ggml/src/ggml-cuda/common.cuh index 8f3d2a26..a0cdab28 100644 --- a/ggml/src/ggml-cuda/common.cuh +++ b/ggml/src/ggml-cuda/common.cuh @@ -579,6 +579,20 @@ struct ggml_cuda_type_traits<GGML_TYPE_IQ2_KT> { }; template<> +struct ggml_cuda_type_traits<GGML_TYPE_IQ3_KT> { + static constexpr int qk = QK_K; + static constexpr int qr = QR4_XS; + static constexpr int qi = QI4_XS; +}; + +template<> +struct ggml_cuda_type_traits<GGML_TYPE_IQ4_KT> { + static constexpr int qk = QK_K; + static constexpr int qr = QR4_XS; + static constexpr int qi = QI4_XS; +}; + +template<> struct ggml_cuda_type_traits<GGML_TYPE_IQ3_K> { static constexpr int qk = QK_K; static constexpr int qr = QR4_XS; |